Hyper recombination accelerates replicative senescence and may promote premature aging

R. Tanner Hagelstroma,b, Krastan B. Blagoevc,d, Laura J. Niedernhofere, Edwin H. Goodwinf, and Susan M. Baileya,1

aDepartment of Environmental and Radiological Health Sciences, Colorado State University, Fort Collins, CO 80523-1618; bPharmaceutical Genomics Division, Translational Genomics Research Institute, Phoenix, AZ 85004; cNational Science Foundation, Arlington, VA 22230; dDepartment of Physics Cavendish Laboratory, Cambridge University, Cambridge CB3 0HE, United Kingdom; eDepartment of Microbiology and Molecular , University of Pittsburgh, School of Medicine and Institute, Pittsburgh, PA 15261; and fKromaTiD Inc., Fort Collins, CO 80524

Edited* by José N. Onuchic, University of California San Diego, La Jolla, CA, and approved August 3, 2010 (received for review May 7, 2010)

Werner syndrome and result from defects in the -cycle arrest known as senescence (12). Most human tissues lack RecQ Werner (WRN) and Bloom (BLM), respectively, and sufficient telomerase activity to maintain telomere length through- display premature aging phenotypes. Similarly, XFE progeroid out life, limiting cell division potential. The majority of syndrome results from defects in the ERCC1-XPF DNA repair endo- circumvent this tumor-suppressor mechanism by reactivating telo- . To gain insight into the origin of cellular senescence and merase (13), thus removing telomere shortening as a barrier to human aging, we analyzed the dependence of sister chromatid continuous proliferation. In some situations, a recombination- exchange (SCE) frequencies on location [i.e., genomic (G-SCE) vs. telo- based mechanism known as “alternative lengthening of ” meric (T-SCE) DNA] in primary human fibroblasts deficient in WRN, (ALT) maintains telomere length in the absence of telomerase (14). BLM, or ERCC1-XPF. Consistent with our other studies, we found evi- orientation (CO)-FISH (15) is a strand-specific dence of elevated T-SCE in telomerase-negative but not telomerase- modification of standard FISH capable of providing information positive backgrounds. In telomerase-negative WRN-deficient cells, not obtainable by any other means, including detection of sister T-SCE—but not G-SCE—frequencies were significantly increased com- chromatid exchange (SCE) recombination between telomeres, pared with controls. In contrast, SCE frequencies were significantly events termed T-SCE (16). T-SCE are emerging as important elevated in BLM-deficient cells irrespective of genome location. In features of certain telomerase-negative backgrounds, such as ERCC1-XPF-deficient cells, neither T- nor G-SCE frequencies differed ALT (17) and early embryogenesis before activation of telomer- from controls. A theoretical model was developed that allowed an ase (18). In the context of combined Werner (WRN) and in silico investigation into the cellular consequences of increased telomerase deficiency in the mouse, both central features of hu- T-SCE frequency. The model predicts that in cells with increased man Werner syndrome (WS) pathogenesis, elevated T-SCE was T-SCE, the onset of replicative senescence is dramatically accelerated observed, with higher rates associated with greater immortaliza- “ even though the average rate of telomere loss has not changed. Pre- tion potential (19). Conditional deletion of the protection of ” mature cellular senescence may act as a powerful tumor-suppressor telomeres (Pot)1a single-stranded telomere-binding has mechanism in telomerase-deficient cells with that cause been shown to elicit a DNA damage response at mouse telo- T-SCE levels to rise. Furthermore, T-SCE-driven premature cellular se- meres, as well as aberrant manifested nescencemaybeafactorcontributing to accelerated aging in Werner as increased T-SCE (20). Human POT1 stimulates the RecQ and Bloom syndromes, but not XFE progeroid syndrome. helicases WRN and Bloom (BLM) to unwind telomeric DNA (21). Furthermore, in the absence of WRN, leading- and lagging- T-SCE | WRN | BLM | ERCC1-XPF | Monte Carlo strand DNA synthesis are uncoupled at replication forks, and POT1 is required for efficient replication of C-rich telomere strands (22). Such studies imply a relationship between T-SCE nderstanding the molecular processes that drive cellular rate and proliferative potential. However, the complexity of the Ureplicative senescence is critical to understanding human process—which entails a large number of telomeres, an ongoing longevity. Much attention has focused on telomere length as exchange process throughout colony growth, exponentially in- a determinant of replicative capacity (1). Telomeres are nucleo- creasing cell number, and independent assortment of chromatids fi protein structures composed of species-speci c tandemly re- into daughter cells during cell division—precludes a simple ana- peated, G-rich DNA sequences that cap and protect chromosome lytical approach to understanding this relationship. We therefore ends from nucleolytic attack and degradation (2). Telomeres also turned to theoretical methods in the search for a causal connec- conceal linear chromosome ends from inappropriate attempts at tion between T-SCE and cell proliferation. double-strand break (DSB) repair that might otherwise join Homologous recombination provides the most plausible end-to-end. The preservation of natural chromo- mechanism by which telomeric DNA could be exchanged between some ends and the rejoining of broken DNA ends, both of which sister telomeres, and in the case of T-SCE, homology can be found are essential for preserving genomic integrity, rely on a common at many points on the sister chromatid. Thus, T-SCE need not subset of (3), and both decline with advancing age (4, 5). exchange equal quantities of DNA. In an unequal T-SCE, one The G-rich nature of telomeric DNA renders it susceptible to sister telomere becomes longer at the expense of the other. Ini- G-quadruplex formation, oxidative damage, and alkylation by elec- tially, we suggested that in telomerase-deficient backgrounds trophiles (6). Telomeres are also unique in their com- unequal T-SCE might confer a proliferative growth advantage to position, being bound by the shelterin complex and associated pro- a subset of cells that stochastically acquired longer telomeres, en- teins, their T-loop structure, and their nucleosomal organization (7, 8). Telomeres protect genetic information by providing a buffer against the end-replication problem, but their terminal position at Author contributions: R.T.H., L.J.N., E.H.G., and S.M.B. designed research; R.T.H., K.B.B., L.J.N., the ends of linear DNA poses a special challenge to their own re- and S.M.B. performed research; R.T.H., K.B.B., L.J.N., E.H.G., and S.M.B. analyzed data; and plication (9). Telomerase is a ribonucleoprotein complex consisting R.T.H., K.B.B., L.J.N., E.H.G., and S.M.B. wrote the paper. minimally of RNA template and catalytic reverse transcriptase The authors declare no conflict of interest. subunits responsible for de novo synthesis of telomeric repeats (10). *This Direct Submission article had a prearranged editor. In the absence of telomerase, telomeres shorten with each cell di- 1To whom correspondence should be addressed. E-mail: [email protected]. vision (11). Critically short telomeres fail to form proper protective This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. end structures, which are sensed as DSBs and trigger a permanent 1073/pnas.1006338107/-/DCSupplemental.

15768–15773 | PNAS | September 7, 2010 | vol. 107 | no. 36 www.pnas.org/cgi/doi/10.1073/pnas.1006338107 Downloaded by guest on October 1, 2021 abling their escape from cellular senescence (16). Conversely, those also displayed normal background T-SCE frequencies (0.29 T-SCE cells inheriting shorter telomeres from an unequal exchange would per chromosome) (Fig. 1). be at risk for triggering senescence prematurely. Using a theoretical To evaluate the consequence of WRN and BLM deficiency on model and Monte Carlo simulations (23), we investigated the pos- T-SCE frequencies in the absence of telomerase, we used an RNA sibility that ALT arises from elevated T-SCE levels and concluded interference siRNA strategy to knock down their expression in it was unlikely unless there was an additional mechanism, such as normal human primary fibroblasts (telomerase-negative). The has been proposed by others (14, 24), through which longer telo- WRN-depleted cells displayed no significant increase in G-SCE meres segregate preferentially into the same daughter cell. This frequencies vs. the mock control (0.12 vs. 0.11) (Fig. 2), although study (23) strongly suggested that in ALT cells, T-SCE coexists with a statistically significant (P < 0.05) increase in T-SCE was ob- served (0.81 vs. 0.25) (Fig. 2). These results are consistent with our an as yet unknown telomere-lengthening mechanism. − − − − SCE is a relatively common event in normal human cells that previous study using cells derived from Wrn / ,Terc / double- occurs along the entire length of a chromosome. Although rec- knockout mice (19). ognized for over 50 y (25), their biological significance remains neither fully understood nor appreciated. SCE are induced by BLM Suppresses SCE Recombination Globally. In contrast to WS lym- DNA-damaging agents that block DNA replication (26), and can phoblasts, human EBV-immortalized BS lymphoblasts (telomerase- lead to deletions, translocations, and loss of heterozygosity (27), positive) displayed a dramatic elevation of G-SCE (0.61) (Fig. 1). manifestations of genomic instability and precursors of carcino- Similar to WS lymploblasts, BS lymphoblasts displayed background genesis. These adverse outcomes likely originate from an inability T-SCE frequencies (0.23) (Fig. 1). Shaded regions on the graph in to resolve replication forks stalled at sites of DNA damage, Fig. 1 represent our established normal human background levels whereupon SCE provides a means to “bypass” the damage (28). for both G- and T-SCE. Interestingly, per unit length of DNA, T-SCE are much more We hypothesized that BLM may perform a reciprocal function frequent than genomic (G)-SCE (16, 29, 30). to WRN, in that BLM may be restricted to suppressing G-SCE, We investigated the contribution of three enzymes, each but WRN involvement is in suppressing T-SCE. However, siRNA fi colocalizing with telomeric DNA and linked to accelerated aging knockdown of BLM in normal human broblasts (telomerase- and increased cancer risk, to regulation of T- and G-SCE fre- negative) demonstrated a role for BLM in suppressing both G- quencies. WS is a rare, autosomal recessive premature aging dis- and T-SCE. Depletion of BLM in the absence of telomerase again WRN, revealed a significant increase in G-SCE frequency vs. the mock ease caused by truncation mutations of a member of the fi RecQ helicase family. WRN is the only RecQ helicase to also control (0.26 vs. 0.08) (Fig. 2), as well as a signi cant increase in possess a 3′-to5′-exonuclease domain, the exact function of which T-SCE (0.40 vs. 0.22) (Fig. 2). is unclear, but which may assist in forming “reverse chicken-foot ERCC1-XPF Does not Influence SCE. To evaluate the role of ERCC1- structures” at stalled replication forks (31). XPF in SCE regulation, we used four congenic primary mouse Bloom syndrome (BS) is also a rare, autosomal recessive dis- embryonic fibroblast (MEF) cell lines (telomerase-positive), two ease associated with segmental, or tissue-specific, premature ag- − − of which were wild-type and two Ercc1 / . No significant dif- ing and a high risk of developing a broad spectrum of cancers, ference in G-SCE frequencies was observed between the wild- primarily carcinomas (32). The defective protein in BS, BLM, is −/− type (0.07, 0.12) (Fig. 3) and the Ercc1 (0.07, 0.07) (Fig. 3) also a RecQ helicase. Extraordinarily elevated levels of G-SCE MEFs. T-SCE frequencies were assessed and no significant dif- are diagnostic of the disease and BS cells are exquisitely sensitive ference was detected between wild-type (0.029, 0.036) (Fig. 3) − − to treatment with the interstrand cross-linking agent mitomycin C, and Ercc1 / (0.041, 0.050) (Fig. 3). These results highlighted the supporting the concept that BS patients suffer from an inability to necessity of analyzing ERCC1’s T-SCE role in a telomerase- resolve complex DNA structures that stall DNA replication. The negative background. helicase activities of both WRN and BLM facilitate resolution of We attempted siRNA knockdown of ERCC1 in normal hu- such DNA structures (e.g., crosslinks and G-quadruplexes), thus man fibroblasts (telomerase-negative). Although similar knock- facilitating progression of the replication machinery (33). XPF down of either WRN or BLM was successful, ERCC1 proved to CELL BIOLOGY XFE progeroid syndrome is caused by mutations in , which be problematic. ERCC1 siRNA-transfected human primary result in reduced expression of ERCC1-XPF (34). ERCC1-XPF is fibroblasts ceased growth almost immediately and metaphases an endonuclease required for nucleotide excision repair and re- could not be obtained for analysis. Similarly, deletion of Ercc1 by − pair of DNA interstrand crosslinks (35). In the absence of transfecting Ercc1 /cond mouse embryonic stem cells with a plas- ERCC1-XPF, replication of cross-linked DNA leads to an accu- mid expressing Cre was not achieved in over 1,000 clones mulation of DSBs and increased chromosomal aberrations rather screened. As further testament to the requirement for ERCC1 than elevated SCE. This suggests that although BLM normally ERCC1 for survival, only one patient with a in has been PHYSICS functions to suppress homologous recombination and SCEs, reported to date (37). ERCC1-XPF promotes them in response to at least some types of We sought to evaluate the role of ERCC1 in T-SCE re- replication stress. Of particular interest, ERCC1-XPF colocalizes combination in primary cells from the ERCC1 patient, 165TOR, with the shelterin protein TRF2 and promotes end-fusion of who despite mutations in the , maintained diminished but critically short telomeres (36). detectable ERCC1-XPF nuclease activity, allowing survival until Investigation of cells defective in WRN, BLM, and ERCC1-XPF 14 mo of age (37). The 165TOR cells grew poorly, even in a low- affords a unique opportunity to decipher the contributions of T- and oxygen environment, but sufficient metaphase spreads for T-SCE G-SCE to cellular senescence. Herein we provide evidence that analysis were obtained. The ERCC1-deficient cells did not dis- increased T-SCE frequency can dramatically accelerate cellular play a T-SCE phenotype; background T-SCE frequencies were senescence of telomerase-negative cells, and that this mechanism identical to those of normal human dermal fibroblast (5C) con- likely functions as a potent tumor-suppressor. trols (0.24 vs. 0.24) (Fig. 3). The 165TOR cells did not grow well enough for G-SCE analysis. However, G-SCE levels determined − − Results in Ercc1 / MEFs (Fig. 3) and ES cells (38) revealed no signif- WRN Suppresses SCE Recombination Specifically in Telomeric DNA. icant difference compared with wild type, and G-SCE frequen- Human EBV-immortalized WS lymphoblasts (telomerase-positive) cies are not influenced by telomerase status. did not display elevated levels of G-SCE (0.08 G-SCE per chro- mosome); shaded regions on the graph in Fig. 1 represent our Senescence in Silico. To understand the role T-SCE play in repli- established normal human background levels. WS lymphoblasts cative senescence, we developed a theoretical model of cell pro-

Hagelstrom et al. PNAS | September 7, 2010 | vol. 107 | no. 36 | 15769 Downloaded by guest on October 1, 2021 A B

i. ii. iii.

Fig. 1. G- and T-SCE frequencies in immortalized human WS and BS lymphoblasts (telomerase-positive). (A) G-SCE frequencies were determined using fluorescence-plus-Giemsa staining, and T-SCE frequencies were determined using CO-FISH. The shaded region represents range of typical frequencies seen in other human cell lines. Error bars were calculated using SEM. (B) Illustrative examples of (i) telomere CO-FISH, characteristic strand-specific (single-sided) signals; (ii) T-SCE, CO-FISH telomere signal split between sister chromatids (arrow); and (iii) G-SCE, color switch between sister chromatids (arrow).

liferation in the presence of telomere recombination (23) and (kb), the rate of telomere loss was 100 bases per telomere per studied in silico the dependence of proliferative potential on cell division, and the critical telomere length was 500 bases. For T-SCE rate. Although a simple idealization of the model in which any set of chosen conditions, the simulation was run 64 times cells have one telomere is analytically tractable (39), in our corresponding to 64 cells growing into colonies. The T-SCE rate experiments human cells had 92 telomeres. Because the model is was set to four values between 0.2 and 0.8 T-SCE per chromo- analytically intractable when many telomeres participate in some per cell division. The results were averaged and graphed, exchanges, we studied it using Monte Carlo simulations. In all and are displayed in Fig. 4 as the senescent cell fraction (fraction cases we assumed that a single, critically short telomere triggers of a colony composed of senescent cells) vs. time expressed in senescence, which is consistent with experimental data (40). Two units of cell divisions. For comparison, with a T-SCE rate of zero, possibilities were considered: one in which only specific chromo- all cells in a colony would senesce simultaneously when the initial somes participated in T-SCE, and one in which T-SCE were shortest telomere dropped below the critical length, in this case randomly distributed among all telomeres. In the former case, at after 26 cell divisions. All cells in our simulated colonies even- each cell division, only the chosen telomeres may exchange tually senesced, and in no case did we observe prolonged colony (quenched case), but in the latter case the chromosomes that growth. In fact, for each nonzero T-SCE rate, colony growth exchange telomere material are “chosen” randomly (annealed) ceased well short of the point in which colonies with no T-SCE from the ensemble of all telomeres in the cell. In the quenched would stop expanding. Thus, contrary to our original hypothesis case, if only a few telomeres participate and these are the short (16) and in agreement with our subsequent study (23), T-SCE in telomeres in the cell, the model indicates that increasing T-SCE and of themselves are not able to overcome telomere-driven rates will prolong the proliferative potential of cells growing into colony senescence. The model allowed us to investigate further colonies (23, 24). However, T-SCE rates observed in this study are the role of telomere number on colony growth, something that sufficiently high so as to indicate that many telomeres participate would be difficult to do experimentally. In cells with few telo- in T-SCE during each cell division, and although not specifically meres, prolonged colony growth is indeed possible, and in the evaluated, there was no indication of preferential participation of special (and biologically irrelevant) case of cells with one telo- a subset of chromosomes. To address this situation we studied the mere per cell, an infinite proliferative potential could be ach- growth of hypothetical cells with 46 chromosomes into colonies. In ieved above a critical T-SCE rate (23, 39). However in a model of the model, the length of each telomere shortened with each cell human cells, to prolong colony growth through unequal T-SCE, division, just as occurs in real cells. Whether or not particular all 92 telomeres of a cell would need to remain above the critical sister telomeres engage in T-SCE is chosen at random, with limit in every division cycle. In contrast, early cell senescence a probability determined by the T-SCE rate at each cell division. requires no more than one telomere to shorten below the critical To simulate unequal T-SCE, the two breakpoints of each ex- limit. Thus, premature replicative senescence would be the pro- change were chosen at random locations within the two sister babilistically favored outcome of unequal T-SCE. telomeres. Each exchange is always balanced such that for a pair With any nonzero T-SCE rate, senescent cells accumulate of sister telomeres, there is no net gain or loss of telomeric DNA. gradually over many cell divisions until a point is reached where Sister chromatids are randomly segregated into daughter cells. If the colony possesses only senescent cells. Comparing the curves in a daughter cell inherited or, because of the end-replication Fig. 4 leads to important insight: T-SCE are remarkably effective problem, obtained one or more telomeres shorter than a critical at accelerating cellular replicative senescence. The implication is length, the cell “senesced.” A senescent cell contributed to the that WS and BS cells, with their high T-SCE rates, will have number of cells in a colony but could not undergo further cell a significantly reduced capacity for colony growth. Therefore, the divisions. Through each simulated cell division, we kept track of model predicts WRN cells will enter senescence with longer av- the number of cells in a colony, whether each cell was senescent or erage telomere length than normal cells. This prediction is sup- not, and the telomere lengths of all chromosome ends. ported by experimental data (41). The smoothness of the coherent Before running a simulation, a set of initial conditions was time evolution as we vary T-SCE levels stems from the large chosen as follows: all telomere lengths were initially 3-kg bases number of cells that were used in the simulations. Fluctuations in

15770 | www.pnas.org/cgi/doi/10.1073/pnas.1006338107 Hagelstrom et al. Downloaded by guest on October 1, 2021 Fig. 2. Small interfering RNA knockdown of WRN and BLM in normal human fibroblasts (telomerase-negative). (A) Western blot analyses of protein levels demonstrate efficient knock- down following siRNA transfection. (B) Evalua- tion of G- and T-SCE frequencies in mock- and siRNA-transfected cells revealed elevated T-SCE upon depletion of either WRN or BLM, although only BLM-depletion resulted in an elevation of G- SCE frequency.

the Monte Carlo results (SEM) are expected to be on the order more likely, that the two RecQ helicases act on different DNA of the inverse square root of the product of the number of cells lesions/structures (e.g., cross-links vs. G-quadruplexes). It is also and the number of telomeres in each cell, a quantity that is neg- possible that strand-specific mechanisms deal with DNA lesions ligible. Further studies will quantify more precisely the influence dependent on whether they reside on the strand replicated by of biological factors that determine the efficacy of the ATM/ leading- or lagging-strand synthesis (22, 43). response to critically short telomeres, as they may be crucial for The separate roles WRN and BLM play in resolving stalled understanding the susceptibility of WS patients to cancer. replication forks may depend on the nature and position of the Just how profoundly colony growth may be constrained by initiating lesion. If the replication fork arrests because of a lesion T-SCE is illustrated in Fig. 4, where the total numbers of cells in or structure occurring in the G-rich telomeric strand (such as an expanding colony are plotted vs. the number of cell divisions a G-quadraplex structure), a 3′ end would be generated, a sub- for several T-SCE rates. If there were no T-SCE, a colony would strate suitable for initiating RAD51-dependent strand invasion. grow to 67 million cells in 26 cell divisions. In contrast, over the BLM is known to associate with RAD51 (44), so may function in 4-fold range of T-SCE rates in our simulation, the average assisting strand-specific strand invasion and resolution of the maximum colony size is reduced from more than one order of resulting . magnitude to nearly three orders of magnitude. In contrast, a replication-fork stalling lesion or structure oc- ′ curring on the C-rich telomeric strand would generate a 5 end, an CELL BIOLOGY Discussion unsuitable substrate for initiation of strand invasion. Therefore, Study of accelerated aging phenotypes affords insight into normal other mechanisms must be enlisted for resolution, such as for- human aging processes, as well as the age-related increase in mation of what has been termed a reverse chicken foot, in which cancer incidence. Because of the intimate relationship between newly synthesized DNA can collapse and use itself as a template aging and cancer, monitoring the dynamic nature of telomere for replication. After which, it can reverse, bypassing the lesion or function can provide valuable biomarkers of underlying determi- prohibitive structure, and replication can proceed (28). It has nants of both cellular aging and . We interrogated PHYSICS telomeric recombination in the premature aging syndromes WS been proposed that the exonuclease domain on WRN, not found and BS and in the recently identified progeroid-like syndrome in any other RecQ helicase, may assist in this processing (45). An fi important aspect of this model is that replication of the strands XFE associated with ERCC1-XPF de ciency (34). Theory and — ’ experiment were combined to relate the obtained recombination can be uncoupled that is, one strand s replication can proceed if — rates to the proliferative capacity of human cells. the other is stalled allowing both strand invasion and reverse We previously observed elevated levels of T-SCE in a murine chicken-foot structure formation to occur. model of combined Wrn and Terc knockout (42), in which the Examination of cells from the only known human patient pre- context of a telomerase-negative background was required to re- senting with accelerated aging associated with ERCC1-XPF fi fl produce a classic human WS phenotype (19). Here, these studies de ciency revealed no in uence on recombination in either were extended in both murine gene knockout and normal human telomeric or genomic DNA (i.e., T- and G-SCE frequencies were fibroblast gene-expression knockdown models. WRN suppressed not affected in this background). Either the residual level of recombination specifically in telomeric DNA, but BLM played ERCC1-XPF activity in the individual was sufficient to circum- a more global role in suppressing SCE recombination. BLM and vent an elevated SCE phenotype, or ERCC1-XPF is not a sup- WRN likely function to restrain T-SCE by resolving stalled rep- pressor of T-SCE and may in fact be required for their formation. lication forks, which occur frequently within telomeric DNA; in- Regardless, the premature aging associated with this case is not dividually, this restraint is incomplete, as either protein alone does likely to be mediated by telomere hyper recombination. Instead, not fully compensate for loss of the other. We speculate that either premature aging might have been caused by nontelomere-related there is an insufficient amount of BLM at telomeres, or perhaps mechanisms of replicative senescence, or by depletion of telomere

Hagelstrom et al. PNAS | September 7, 2010 | vol. 107 | no. 36 | 15771 Downloaded by guest on October 1, 2021 A

B

Fig. 3. G- and T-SCE frequencies with Ercc1 deficiency. (A)G-andT-SCEfre- − − quencies in two independent Ercc1 / and congenic wild-type primary MEF lines (telomerase-positive). No significant difference in G- or T-SCE frequencies be- tween the two genotypes was observed. (B) T-SCE frequencies in ERCC1-deficient human patient primary fibroblasts (165TOR). Even in this telomerase-negative background, no difference in T-SCE frequencies was observed as compared with normal human fibroblasts. G-SCE frequencies were not determined.

reserves brought on by an elevated rate of cell turnover to com- pensate for frequent cell death. Fig. 4. Senescence in silico. (A) T-SCE accelerate colony senescence. Com- A WRN cell culture can be thought of as a collection of puter simulations were performed in which colonies grew from single cells. expanding colonies, each burdened by a type of telomere in- The four curves correspond to T-SCE levels covering a range from 0.2 to 0.8 stability that drives the culture as a whole toward accelerated se- T-SCE per chromosome per cell division. As T-SCE rates rise, so does the nescence. Computer simulations suggest that higher T-SCE rates fraction of the colony composed of senescent cells at any cell division. (B) lead to smaller colony sizes after the same number of divisions. T-SCE constrain colony growth. The four curves generated through com- Because senescent cells in a colony are incapable of cell division, puter simulation demonstrate that T-SCE can have a strong negative impact nonsenescent WS cells have to divide more times to repopulate on colony size. With increasing T-SCE rates colonies increase in size more slowly, and the average colony size attained when all cells have senesced a colony in comparison with cells without the WRN mutation, decreases almost exponentially. which almost certainly contributes to the senescence of WS cell cultures after fewer in vitro passages. How might elevated T-SCE contribute to cancer risk and pre- In the quest to identify the cause of human aging, much atten- mature aging? An in vivo study of the tissues of baboons found the tion has focused on replicative senescence driven by telomere loss. proportion of senescent cells increased with age rising to about Our studies strongly implicate T-SCE as a second potent driver of 20% in the oldest animals (46). This study suggests that a sen- replicative senescence. To accurately predict a cell’s replicative fi escent cell fraction of 0.2 may suf ciently disrupt organ function fate, T-SCE evaluation should be included in studies attempting to to cause the effects of advanced old age. Our in silico study implies relate genetics or lifestyle factors to telomere dynamics and aging. this level of replicative senescence would be reached at an earlier chronological age when the T-SCE rate is high. Colony growth Materials and Methods impairment during the growth spurt of the teen years may also Cell Lines. Human WS and BS lymphoblasts were obtained from Coriell (WS- contribute to the small stature of adult WS patients. A causal AG04103 and BS-GM16375, respectively). Both were grown in RPMI 1640 connection to cancer is likely to be more complex. Compared with (HyClone) supplemented with 15% FBS (Sigma-Aldrich) and 1% pen/strep normal individuals of the same age, high T-SCE rates cause WS (HyClone). Low passage normal human dermal fibroblasts (Cascade Biolog- and BS patients to challenge the ATM/p53 guardian barrier with ics #C-004–5C) were maintained in α-MEM medium (HyClone) supplemented − − a larger number of cells having short telomeres. On the other with 15% FBS (Sigma Aldrich) and 1% pen-strep (HyClone). Ercc1 / primary hand, accelerated replicative senescence is likely to terminate MEFs were generated as per ref. 34. Primary human fibroblasts from ERCC1 many premalignant clones before they acquire a full suite of on- patient 165TOR were obtained and maintained as previously described (37). cogenic mutations, thereby suppressing the rate of tumor forma- All cell lines were incubated at 37 °C in an atmosphere of 95% air and 5% tion. With very few parameters, our model can be directly related carbon dioxide. to experiments with cultured cells. The model, however, is limited G-SCE Analysis. Fluorescence-plus-Giemsa was performed according to pro- in its application to the age-related of tissues and tocol of Perry and Wolff with some modification (47). Cells were trypsinized − organisms because it lacks important mechanisms, such as geno- and subcultured into medium containing 1 × 10 5M of the thymidine analog mic instability, reactivation of telomerase, and bypass of check- BrdU (Sigma-Aldrich) for two cell cycles. Colcemid (0.1 mg/mL; Invitrogen) points en route to . In future studies, we was added during the final 3 to 4 h to accumulate mitotic figures. Slides hope to incorporate these mechanisms to gain a better apprecia- were air dried and stained with Hoechst 33258 (0.50 mg/mL; Sigma-Aldrich) tion for the interrelationship between cancer and aging, and their for 15 min, exposed to 365 nm UV light (Stratalinker 2400) for 25 min, then mutual dependence on telomere fitness. stained in 2% Giemsa (Fisher Scientific). Harlequin (differential) staining was

15772 | www.pnas.org/cgi/doi/10.1073/pnas.1006338107 Hagelstrom et al. Downloaded by guest on October 1, 2021 visualized with light microscopy and SCE scored as a “color switch” between reagent/1,000 μL of Opti-MEM), and then added to exponentially growing sister chromatids. cells in culture media without FBS or antibiotics. Cells were incubated for 6 h at 37 °C, then culture media with FBS and pen/strep was re-added. Lystates T-SCE Analysis. CO-FISH was performed as previously described (48) and in SI were collected after 24 to 72 h, dependent on time of maximum knockdown fl Materials and Methods. Brie y, cells were subcultured into medium con- for each protein. taining BrdU (1 × 10−5M; Sigma-Aldrich) for one cell cycle. Slides were stained with Hoechst 33258 (0.50 μg/mL; Sigma-Aldrich) for 15 min and ex- Western Blot Analysis. See SI Materials and Methods for details. The fol- posed to 365 nm UV light (Stratalinker 2400) for 25 min. BrdU-incorporated lowing primary antibodies and their corresponding ratios were used: WRN strands were digested with Exonuclease III (3 U/μL; Promega) for 10 min. A (c-19) antibody (1:1,000; Santa Cruz), BLM(RBX) antibody [1:1,000; Chem- Cy-3 conjugated (TTAGGG)3 PNA telomere probe (0.2 μg/mL; Applied Bio- systems) was hybridized at 37 °C for 1.5 h. Chromosomes were counter- icon/Upstate (Millipore)], actin (1:250; Santa Cruz). The following secondary stained with DAPI (Vectashield, Vector Laboratories). Preparations were antibodies were used for detection: for WRN, donkey anti-goat HRP examined and images captured. T-SCE was scored as a CO-FISH telomere (1:1,000; Santa Cruz); for BLM, goat anti–rabbit HRP (1:1,000; Santa Cruz); signal split between the two chromatids of a metaphase chromosome. for actin, (1:1,000; Santa Cruz).

Statistical Analysis. A minimum of 50 metaphases per sample was scored, ACKNOWLEDGMENTS. We thank Ms. Christine Battaglia for expert technical unless 50 scorable metaphases were not available. Error bars were calculated assistance and critical reading of the manuscript. This research was supported using SEM, and SCE frequencies were calculated on a per chromosome basis. in part by the National Aeronautics and Space Administration (NNJ04HD83G and NNX08AB65G), National Institute of Environmental Health Sciences Grant Small Interfering RNA Transfection. The siRNA sequences (20 nM) (SI Materials ES16114 (to L.J.N.), and Ellison Medical Foundation Grant AG-NS-0303-05 and Methods) were incubated for 30 min in HiPerfect transfection reagent (to L.J.N.). K.B.B. was supported by the National Science Foundation while (Qiagen) and Opti-MEM (Invitrogen) at a 2.5:1,000 ratio (2.5 μL transfection working at the Foundation.

1. Royle NJ, et al. (2009) The role of recombination in telomere length maintenance. 25. Taylor JH (1958) Sister chromatid exchanges in tritium-labeled chromosomes. Biochem Soc Trans 37:589–595. Genetics 43:515–529. 2. de Lange T (2009) How telomeres solve the end-protection problem. Science 326: 26. Nakanishi Y, Schneider EL (1979) In vivo sister-chromatid exchange: A sensitive 948–952. measure of DNA damage. Mutat Res 60:329–337. 3. O’Sullivan RJ, Karlseder J (2010) Telomeres: Protecting chromosomes against genome 27. Martin GM (2005) Genetic modulation of senescent phenotypes in Homo sapiens. Cell instability. Nat Rev Mol Cell Biol 11:171–181. 120:523–532. 4. Harley CB, Futcher AB, Greider CW (1990) Telomeres shorten during ageing of human 28. Wilson DM, 3rd, Thompson LH (2007) Molecular mechanisms of sister-chromatid fibroblasts. Nature 345:458–460. exchange. Mutat Res 616:11–23. 5. Sedelnikova OA, et al. (2008) Delayed kinetics of DNA double-strand break processing 29. Cornforth MN, Eberle RL (2001) Termini of human chromosomes display elevated in normal and pathological aging. Aging Cell 7:89–100. rates of mitotic recombination. Mutagenesis 16:85–89. 6. von Zglinicki T (2000) Role of oxidative stress in telomere length regulation and 30. Rudd MK, et al. (2007) Elevated rates of sister chromatid exchange at chromosome replicative senescence. Ann N Y Acad Sci 908:99–110. ends. PLoS Genet 3:e32. 7. de Lange T (2005) Shelterin: The protein complex that shapes and safeguards human 31. Rossi ML, Ghosh AK, Bohr VA (2010) Roles of Werner syndrome protein in protection telomeres. Dev 19:2100–2110. of genome integrity. DNA Repair (Amst) 9:331–344. 8. Pisano S, Galati A, Cacchione S (2008) Telomeric nucleosomes: Forgotten players at 32. Hickson ID (2003) RecQ helicases: Caretakers of the genome. Nat Rev Cancer 3: chromosome ends. Cell Mol Life Sci 65:3553–3563. 169–178. 9. Verdun RE, Karlseder J (2007) Replication and protection of telomeres. Nature 447: 33. Wu L, Hickson ID (2003) The Bloom’s syndrome helicase suppresses crossing over 924–931. during homologous recombination. Nature 426:870–874. 10. Greider CW, Blackburn EH (2004) Tracking telomerase. Cell 116(2 Suppl):S83–S86, 1 p 34. Niedernhofer LJ, et al. (2006) A new progeroid syndrome reveals that genotoxic stress following S86. suppresses the somatotroph axis. Nature 444:1038–1043. 11. Shore D, Bianchi A (2009) Telomere length regulation: Coupling DNA end processing 35. Niedernhofer LJ, et al. (2004) The structure-specific endonuclease Ercc1-Xpf is to feedback regulation of telomerase. EMBO J 28:2309–2322. required to resolve DNA interstrand cross-link-induced double-strand breaks. Mol Cell 12. Harley CB, Vaziri H, Counter CM, Allsopp RC (1992) The telomere hypothesis of Biol 24:5776–5787. cellular aging. Exp Gerontol 27:375–382. 36. Zhu XD, et al. (2003) ERCC1/XPF removes the 3′ overhang from uncapped telomeres 13. Kim NW, et al. (1994) Specific association of human telomerase activity with immortal and represses formation of telomeric DNA-containing double minute chromosomes. cells and cancer. Science 266:2011–2015. Mol Cell 12:1489–1498. fi

14. Cesare AJ, Reddel RR (2010) Alternative lengthening of telomeres: Models, 37. Jaspers NG, et al. (2007) First reported patient with human ERCC1 de ciency has CELL BIOLOGY mechanisms and implications. Nat Rev Genet 11:319–330. cerebro-oculo-facio-skeletal syndrome with a mild defect in nucleotide excision repair 15. Bailey SM, Goodwin EH, Cornforth MN (2004) Strand-specific fluorescence in situ and severe developmental failure. Am J Hum Genet 80:457–466. hybridization: The CO-FISH family. Cytogenet Genome Res 107:14–17. 38. Niedernhofer LJ, et al. (2001) The structure-specific endonuclease Ercc1-Xpf is 16. Bailey SM, Brenneman MA, Goodwin EH (2004) Frequent recombination in telomeric required for targeted gene replacement in embryonic stem cells. EMBO J 20: DNA may extend the proliferative life of telomerase-negative cells. Nucleic Acids Res 6540–6549. 32:3743–3751. 39. Antal T, Blagoev KB, Trugman SA, Redner S (2007) Aging and immortality in a cell 17. Bechter OE, Zou Y, Shay JW, Wright WE (2003) Homologous recombination in human proliferation model. J Theor Biol 248:411–417. telomerase-positive and ALT cells occurs with the same frequency. EMBO Rep 4: 40. Hemann MT, Strong MA, Hao LY, Greider CW (2001) The shortest telomere, not

1138–1143. average telomere length, is critical for cell viability and chromosome stability. Cell PHYSICS 18. Liu L, et al. (2007) Telomere lengthening early in development. Nat Cell Biol 9: 107:67–77. 1436–1441. 41. Schulz VP, et al. (1996) Accelerated loss of telomeric repeats may not explain 19. Laud PR, et al. (2005) Elevated telomere-telomere recombination in WRN-deficient, accelerated replicative decline of Werner syndrome cells. Hum Genet 97:750–754. telomere dysfunctional cells promotes escape from senescence and engagement of 42. Chang S, et al. (2004) Essential role of limiting telomeres in the pathogenesis of the ALT pathway. Genes Dev 19:2560–2570. Werner syndrome. Nat Genet 36:877–882. 20. Wu L, et al. (2006) Pot1 deficiency initiates DNA damage checkpoint activation and 43. Crabbe L, Verdun RE, Haggblom CI, Karlseder J (2004) Defective telomere lagging aberrant homologous recombination at telomeres. Cell 126:49–62. strand synthesis in cells lacking WRN helicase activity. Science 306:1951–1953. 21. Opresko PL, et al. (2005) POT1 stimulates RecQ helicases WRN and BLM to unwind 44. Lahkim Bennani-Belhaj K, et al. (2010) The limits the telomeric DNA substrates. J Biol Chem 280:32069–32080. lethality associated with RAD51 deficiency. Mol Cancer Res 8:385–394. 22. Arnoult N, Saintome C, Ourliac-Garnier I, Riou JF, Londoño-Vallejo A (2009) Human 45. Sowd G, Wang H, Pretto D, Chazin WJ, Opresko PL (2009) stimulates POT1 is required for efficient telomere C-rich strand replication in the absence of the Werner syndrome protein branch migration activity. J Biol Chem 284:34682–34691. WRN. Genes Dev 23:2915–2924. 46. Herbig U, Ferreira M, Condel L, Carey D, Sedivy JM (2006) Cellular senescence in aging 23. Blagoev KB, Goodwin EH (2008) Telomere exchange and asymmetric segregation of primates. Science 311:1257. chromosomes can account for the unlimited proliferative potential of ALT cell 47. Perry P, Wolff S (1974) New Giemsa method for the differential staining of sister populations. DNA Repair (Amst) 7:199–204. chromatids. Nature 251:156–158. 24. Muntoni A, Reddel RR (2005) The first molecular details of ALT in human tumor cells. 48. Bailey SM, Cornforth MN, Kurimasa A, Chen DJ, Goodwin EH (2001) Strand-specific Hum Mol Genet 14 Spec No. 2: R191–R196. postreplicative processing of mammalian telomeres. Science 293:2462–2465.

Hagelstrom et al. PNAS | September 7, 2010 | vol. 107 | no. 36 | 15773 Downloaded by guest on October 1, 2021